You are on page 1of 10

Engineering Structures 33 (2011) 245–254

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Performance of axially restrained concrete encased steel composite columns at


elevated temperatures
Ben Young a , Ehab Ellobody b,∗
a
Department of Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong
b
Department of Structural Engineering, Faculty of Engineering, Tanta University, Tanta, Egypt

article info abstract


Article history: The structural performance of axially restrained concrete encased steel composite columns at elevated
Received 22 January 2010 temperatures is investigated in this study. An efficient nonlinear 3-D finite element model was presented
Received in revised form for the analysis of the pin-ended axially loaded columns. The restraint ratios varied from 20% to 100%
20 July 2010
of the axial stiffness of the composite columns at ambient temperature. The finite element model was
Accepted 14 October 2010
Available online 9 November 2010
verified against published test results on axially restrained concrete encased steel composite columns
at elevated temperatures. The columns investigated had different cross-sectional dimensions, different
Keywords:
coarse aggregates and different load ratios during fire. The nonlinear material properties of steel, concrete,
Axially restrained longitudinal and transverse reinforcement bars as well as the effect of concrete confinement at ambient
Concrete encased steel and elevated temperatures were considered in the finite element model. The interface between the
Composite columns steel section and concrete, the longitudinal and transverse reinforcement bars, and the reinforcement
Elevated temperatures bars and concrete were also considered allowing the bond behaviour to be modelled and the different
Finite element components to retain its profile during the deformation of the column. The initial overall geometric
Modelling imperfection was carefully included in the model. The time–temperature relationships, deformed shapes
Fire resistance at failure, time–axial displacement relationships, failure modes and fire resistances of the columns were
Structural fire design
evaluated by the finite element model and compared well against test results. Furthermore, the variables
that influence the fire resistance and behaviour of the axially restrained composite columns comprising
different axial restraint ratios, different load ratios during fire, different coarse aggregates and different
slenderness ratios were investigated in a parametric study. It is shown that axially restrained composite
columns behave differently in fire compared to the unrestrained columns since the typical ‘‘runaway’’
failure was not predicted from the finite element analysis. The fire resistances of the composite columns
obtained from the finite element analysis were compared with the design values obtained from the
Eurocode 4 for composite columns at elevated temperatures. It is shown that the EC4 is generally
conservative for all the axially restrained concrete encased steel composite columns, except for some
columns with higher load and slenderness ratios.
© 2010 Elsevier Ltd. All rights reserved.

1. Introduction In addition, fire tests were found in the literature on unrestrained


concrete-encased steel composite columns as detailed in [11–13].
Concrete encased steel composite columns are gaining popu- Looking at a whole building, the behaviour of a heated column
larity in tall building construction due to their high strength, full under fire conditions is affected by the adjacent cool structural
usage of materials, high stiffness and ductility, toughness against members. The cool structural elements would act as a restraint
seismic loads, significant savings in construction time and mainly to thermal expansion. The heated column interacts with the
good in fire resistance. Experimental investigations were con- adjacent structural members depending on the restraint’s stiffness
ducted at ambient temperature on unrestrained concrete encased and direction. Hence, when a whole building is exposed to
steel composite columns as detailed in [1–3]. Furthermore, ana- fire the common is the restrained condition. However, studying
lytical studies at ambient temperature on unrestrained concrete retrained structural elements under fire conditions requires special
care to apply restraining actions and to maintain these actions
encased steel composite columns have been performed in [4–10].
during fire tests. Hence, only very limited number of tests were
found in the literature investigating the behaviour of axially
restrained concrete encased steel composite columns at elevated
∗ Corresponding author. Tel.: +20 40 3315860; fax: +20 40 3315860. temperatures, leading to the present investigation. The fire tests
E-mail address: ehab_ellobody@tanta.edu.eg (E. Ellobody). on axially restrained concrete encased I-section steel composite
0141-0296/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2010.10.019
246 B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254

conducted using the finite element program [16]. The effects of


Nomenclature initial overall geometric imperfections were included in the finite
element model using different values of half-sign curves. The
Ac Cross-sectional area of concrete assumptions used in [12–14] have resulted in inaccurate results
Ar Cross-sectional area of reinforcement bars and shortcomings in the analyses, as acknowledged by the authors
As Cross-sectional area of structural steel of [12–14]. In addition, the unit of Young’s modulus of concrete
B Overall width of cross-section (smaller dimension) was mistakenly calculated in [12,14] which resulted in a lower
b Width of structural steel section estimated design capacity for the columns in cold conditions and
c1 Cover to structural steel consequently inaccurate fire loads and restraint ratios, leading to
c2 Cover to structural steel the present investigation.
D Overall depth of cross-section (larger dimension) Most of the shortcomings reported in [12–14] were eliminated
d Depth of structural steel section in recent finite element models on concrete encased steel
Ec20 Elastic modulus of concrete at ambient temperature composite columns at ambient and elevated temperatures as
Er20 Elastic modulus of reinforcement bars at ambient detailed in [17,18], respectively. The models [17,18] were focused
temperature on axially unrestrained concrete encased steel composite columns.
Es20 Elastic modulus of structural steel at ambient The models were nonlinear 3-D models considering the inelastic
temperature material properties of steel, concrete, longitudinal and transverse
e Emissivity of concrete reinforcement bars as well as the effect of concrete confinement
F Flexural buckling failure mode and concrete cracking. The interface between the steel section and
FE Finite element concrete, the longitudinal and transverse reinforcement bars, and
fc Unconfined compressive cylinder strength of con- the reinforcement bars and concrete were included in the models.
crete The initial overall geometric imperfection was also considered in
fy Yield stress the models. However, the behaviour of axially restrained concrete
fys Yield stress of structural steel encased steel composite columns exposed to fire is quite different
fyr Yield stress of reinforcement bars from that of unrestrained columns as presented in [14]. Restraining
fu Ultimate stress axial expansion results in different time–axial displacements,
fur Ultimate stress of reinforcement bars different failure modes and different fire resistances than that
fus Ultimate stress of structural steel of unrestrained columns, which is addressed in the current
K 20 Axial stiffness of composite column at ambient investigation.
temperature The main objective of this study is to present an efficient
k Effective length factor nonlinear 3-D finite element model for the analysis of axially
ks Applied spring stiffness restrained pin-ended concrete encased steel composite columns
L Length of column at elevated temperatures. The finite element model considered
Le Effective length of column different axial restraint ratios, the inelastic behaviour of the
PEC4 Unfactored design strength calculated according to composite column components, the interfaces between these
EC4 components and concrete confinement at elevated temperatures.
PFE Column strength obtained from finite element The model was built in ABAQUS [19] and also validated against
analysis published experimental results at elevated temperatures [14]. The
PFiret Load during fire initial overall geometric imperfections were carefully included
S Transverse reinforcement spacing in the finite element model. A parametric study was performed
us1 Reinforcement bar dimension to investigate the effects of different axial restraint ratios, load
us2 Reinforcement bar dimension ratios during fire, different coarse aggregate types and different
λ Relative slenderness ratio slenderness ratios on the fire resistance and behaviour of the
ρs Structural steel cross-section area to full section axially restrained composite columns. The fire resistances of
area ratio the composite columns were compared with the design values
αc Convective coefficient calculated using EC4 [20].
β Restraint ratio
2. Finite element modelling of restrained columns at elevated
temperatures
columns were conducted by Huang et al. [14]. The tests [14]
comprised three fire tests having different axial restraint ratios. 2.1. Summary of finite element modelling of unrestrained columns
The tests provided valuable data in the form of time–temperature
curves of the reinforcement, concrete and encased steel section as Fig. 1 shows the concrete encased steel composite column
well as time–axial displacement behaviour and test periods that section investigated in this study. The composite column consists
could be used in the verification of the finite element model. of the steel section, longitudinal reinforcement bars, transverse
Numerical models were developed on concrete encased steel reinforcement bars and concrete. The unrestrained columns were
composite columns at elevated temperature for unrestrained previously modelled at ambient and elevated temperatures
columns [11–13] and for restrained columns [14]. However, these as detailed in [17,18], respectively. In the modelling of the
models were based on assumptions including the bonding among unrestrained columns at ambient temperature [17], the composite
concrete, rebar and steel surface that assumed neglecting slippage column components were modelled using a combination of 3-
and cracking, no concrete spalling, no concrete confinement, D eight-node and six-node solid elements (C3D8 and C3D6). The
no transverse reinforcement and no water movement effects elements have three degrees of freedom per node and capable
in the analysis. The numerical studies [12–14] predicted the of modelling all the composite column components at ambient
temperatures in the composite cross-section using a 2-D model and elevated temperatures. Concrete was modelled using the
developed by the finite element package [15]. The temperatures damaged plasticity model. The concrete damaged plasticity model
predicted were then included in a structural–thermal analysis uses the concept of isotropic damaged elasticity, in combination
B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254 247

Table 1
Specimen dimensions and material properties of concrete encased steel composite columns exposed to fire.
Test Restraint condition Dimensions Structural steel section Material properties Refs.
B (mm) D (mm) Section As (mm2 ) ρs (%) fc (MPa) fys (MPa) fur (MPa)

RCC02 Restrained 300 300 UC 152 × 152 × 37 4710 5.23 43 320 460 [14]
RCC03 Restrained 300 300 UC 152 × 152 × 37 4710 5.23 43 320 460 [14]
RCC04 Restrained 300 300 UC 152 × 152 × 37 4710 5.23 43 320 460 [14]
RCC01 Unrestrained 300 300 UC 152 × 152 × 37 4710 5.23 43 320 460 [14]
SZCC02 Unrestrained 300 300 UC 203 × 203 × 52 6630 7.37 43 317 460 [12,13]

Table 2
Comparison between test and finite element results.
Test [Refs.] Restraint condition Test period (min) FE period (min) FE/test
FE period/test period

RCC02 [14] Restrained 417 367 0.88


RCC03 [14] Restrained 422 368 0.87
RCC04 [14] Restrained 366 376 1.03
RCC01 [14] Unrestrained 414 372 0.90
SZCC02 [12,13] Unrestrained 229 232 1.01

Mean 0.94
COV 0.081

static load. Eigenvalue buckling analysis for the composite column


was conducted to predict the first buckling mode, which is then
factored by the magnitude of the initial overall geometric imper-
fection and inserted in the model. The magnitude of the initial
overall geometric imperfection was taken as L/1000, where L is the
length of the pin-ended column. The time–temperature relation-
ships, time–axial displacement relationships, deformed shapes at
failure, failure modes and composite column fire resistances have
been predicted using the finite element model and compared well
against the test results. For comparison purposes in this study, the
unrestrained concrete encased steel composite columns tested at
elevated temperatures were modelled adopting the same approach
in [18] and the results are presented in Tables 1–3. Fig. 2 shows the
Fig. 1. Cross-section of restrained concrete encased steel composite column.
finite element model of the restrained concrete encased steel com-
posite columns investigated in this study.
with isotropic tensile and compressive plasticity, to represent
the inelastic behaviour of concrete. The model assumes that
the uniaxial tensile and compressive response of concrete is 2.2. Finite element modelling of restrained columns
characterized by damaged plasticity. Under uniaxial compression
the response is linear until the value of proportional limit stress is The axially restrained concrete encased steel composite columns
reached. Under uniaxial tension the stress–strain response follows investigated in this study are shown in Fig. 2. The composite
a linear elastic relationship until the value of the failure tensile columns were modelled adopting the same approach presented by
stress is reached. The tensile failure stress was assumed to be Ellobody and Young [17,18]. The composite column was divided
0.1 times the compressive strength of concrete. The softening into the highly confined concrete, the partially confined concrete
stress–strain response, past the maximum tensile stress, was and the unconfined concrete zones, as shown in Fig. 2. The confine-
represented by a linear line defined by the fracture energy and ment factors for the highly and partially confined concrete zones
crack band width. The fracture energy Gf (energy required to open were predicted as given in [10]. The confinement zones were sim-
a unit area of crack) was taken as 0.12 N/mm. The fracture energy plified, as adopted by Mirza and Skrabek [1], by taking the highly
divided by the crack band width (h) was used to define the area confined concrete from the web of the steel section to the mid-
under the softening branch of the tension part of the stress–strain flange, the partially confined concrete from the mid-flange to the
the reinforcement steel at ambient temperature given in the centerlines of the longitudinal reinforcement and finally, the un-
EC3 [21] and EC2 [22], respectively, were adopted with measured confined concrete were the remaining external zone as shown in
values of the yield stress (fy ) and ultimate stress (fu ) used in Fig. 2. The pin-end conditions were modelled by allowing the sup-
the tests. The composite column strengths, axial shortening at port end plates to rotate about the axis of the plate, direction 2–2
failure and failure modes of the columns have been predicted using in Fig. 2, which is bending about the minor axis. The upper and
the finite element model and compared well against published lower end plate nodes were connected to the matching specimen
experimental results. end nodes. The center point of the outside surface of the bottom
In the modelling of unrestrained concrete encased steel com- plate was restrained against all degrees of freedom while that of
posite columns at elevated temperatures [18], an uncoupled the top plate, the loading position, was allowed to displace in the
thermal–structural analysis was performed. The temperature dis- vertical direction only, direction 3–3 in Fig. 2. The load was applied
tribution in the columns was predicted from the thermal anal- as a static concentrated load.
ysis based on the measured temperature from the tests [12,13] The steel section–concrete, the longitudinal reinforcement
and the standard fire curve given in the EC1 [23]. The predicted bar–lateral reinforcement bar, and the reinforcement bar–concrete
temperatures from the thermal analysis were inserted in the interfaces were modelled by interface elements. The interface
structural analysis. The columns were heated under a constant elements require defining two surfaces that are the master and
248 B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254

(a) Column specimen. (b) Restraint to column specimen.

Fig. 2. Finite element model of restrained concrete encased steel composite columns.

Table 3
Fire resistances of the tests predicted using the standard fire curve.
Test [Refs.] Restraint condition Ambient temperature Elevated temperature
λ FE EC4 PFE /PEC4 Fire load Load ratio FE
PFE (kN) PEC4 (kN) PFire (kN) PFire /PEC4 Fire resistance (min)

RCC02 [14] Restrained 0.56 3966 3922 1.01 1106 0.28 143
RCC03 [14] Restrained 0.56 3966 3922 1.01 1106 0.28 142
RCC04 [14] Restrained 0.56 3966 3922 1.01 1106 0.28 148
RCC01 [14] Unrestrained 0.56 3966 3922 1.01 1106 0.28 137
SZCC02 [12,13] Unrestrained 0.56 4544 4351 1.04 1735 0.40 103
Mean – – – – 1.02 – – –
COV – – – – 0.013 – – –

slave surfaces. The master surface within this model is defined same convective coefficients and concrete emissivity value were
as the concrete surface surrounding the steel section and the previously used, with reasonable accuracy, as detailed in [26]. The
reinforcement bars that are the slave surfaces. The interface stress–strain–temperature curves for concrete under compression
elements are formed between the master and slave surfaces and and tension were based on the reduction factors given in EC4 [27].
monitor the displacement of the slave surface in relation to the The influence of temperature on the concrete model was previ-
master surface. When the two surfaces remain in contact, the ously detailed in [26]. The specific heat and thermal conductivity of
slave surface can displace relative to the master surface based concrete were calculated according to EC2 [28] with the measured
on the coefficient of friction between the two surfaces, which is moisture content considered in the calculation of the specific heat
taken as 0.25 [24,25]. When the two surfaces are in contact, the of concrete. Concrete spalling was not considered in the present
forces normal to the master surface can be transmitted between analysis. The stress–strain curves for the structural steel and re-
the two surfaces. When the two surfaces separate, the relative inforcement bars at elevated temperatures were calculated based
displacement between the two surfaces can still be monitored but on the reduction factors given in the EC4 [20]. The specific heat,
the forces normal to the master surface cannot be transmitted. thermal conductivity and thermal expansion of the structural steel
However, the two surfaces cannot penetrate each other. and reinforcement bars were also calculated according to EC4 [20].
In this study, a similar finite element modelling technique of Thermal expansion coefficients for concrete with different aggre-
the unrestrained concrete encased steel composite columns at gates were based on linear thermal expansion coefficients recom-
elevated temperatures [18] was used for the axially restrained mended by Klieger and Lamond [29]. Following the heat transfer
columns. A thermal heat transfer analysis was performed using the analysis, a structural–thermal analysis was performed for the axi-
modified experimental curve [14] together with the standard fire ally restrained columns.
curve given in the EC1 [23], as shown in Fig. 3. The temperature The axial restraint was applied to the finite element model
distribution in the axially restrained columns was predicted from using spring elements. The spring had linear elastic stiffness (ks )
the heat transfer analysis. A constant convective coefficient (αc ) of and the initial separation between the restraining beam and the
25 W/m2 K was assumed for the exposed surface and 9 W/m2 K specimen [14] was considered in the finite element analysis. The
was assumed for the unexposed surface. The radiative heat flux applied spring stiffness ks was taken as a ratio (β ) of the axial
was calculated using a concrete emissivity (e) value of 0.7. The stiffness of the composite column at ambient temperature (K 20 )
B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254 249

determined from Eq. (1).


K 20 = [Es20 As + Er20 Ar + Ec20 Ac ]/Le (1)
where, Es20 , Er20 and Ec20 are the elastic moduli of elasticity for struc-
tural steel, reinforcement bars and concrete at ambient temper-
ature, respectively, and As , Ar , Ac are the cross-sectional area of
structural steel, reinforcement bars and concrete, respectively, and
Le is the effective length of the pin-ended column.

3. Summary of fire tests on axially restrained composite


columns

The experimental investigation conducted by Huang et al. [14]


on axially restrained concrete encased steel composite columns at
elevated temperatures was used in the verification of the finite
element model built in ABAQUS [19] in this study. The columns
were pin-ended axially loaded columns. Table 1 summarizes
the measured dimensions and material properties of the fire Fig. 3. Time–temperature curves used in the heat transfer analysis.
tests [14]. The measured dimensions and material properties of the
unrestrained concrete encased steel composite column conducted temperatures. Fig. 4 plotted the time–temperature relationships at
in [13,14] are also shown in Table 1. The fire tests detailed the mid-height of column RCC03. The relationships were predicted
in [14] comprised three axially restrained concrete encased steel at the position of the longitudinal reinforcement bar, at the mid-
composite columns labeled RCC02, RCC03 and RCC04 and one
web point of the structural steel and at the quarter point along the
reference unrestrained column RCC01 having the same square
symmetry axis 1–1 of the concrete, as shown in Fig. 2. It can be seen
cross-section (B × D) of 300 × 300 mm and the same structural
that the FE model closely predicted the temperatures at different
steel section UC 152 × 152 × 37. Applied loads on the composite
locations of the composite column.
columns were controlled and recorded by a servo-hydraulic
The fire test periods and finite element periods are compared
actuator. Restraint to thermal expansion of the composite columns
in Table 2 for the axially restrained and unrestrained specimens
was applied by a transverse steel beam, simply supported on
detailed in [14]. The unrestrained specimen detailed in [13,14]
to a set of reaction frames as presented in [14]. Different axial
and analyzed in [18] is also presented in Table 2. It can be
restraint stiffness was achieved by choosing different sizes of
shown that a generally good agreement between the experimental
transverse restrained beam placed external to the furnace. The
and numerical results was obtained. The mean value of the FE
axial restraint to the columns RCC02, RCC03 and RCC04 using
period/Test period ratio is 0.94 with the corresponding coefficient
different restraining steel beams provided axial restraint stiffness
ks of 30.61, 27.95 and 52.45 (MN/m), respectively, and restraint of variation (COV) of 0.081, as shown in Table 2. However, it should
ratio (β ) of 0.03, 0.03 and 0.05, respectively, of the composite be noted that the COV may not be representative for this small
column axial stiffness at ambient temperature. The columns had number of tests. The verified finite element model was also used in
a structural steel cross-sectional area/overall area cross-sectional analyzing the composite columns using the standard fire curve [23]
(ρs ) ratio of 5.23%. All the column specimens had a length and the results are summarized in Table 3. First, the column
of 3300 mm. The specimens had an average concrete cylinder strengths (PFE ) were predicted at ambient temperature using the
strength (fc ) of 43 MPa and structural steel yield stresses (fys ) of finite element model presented in [17]. The unfactored design
strengths (PEC4 ) calculated using the EC4 [27] are also presented
320 MPa. The relative slenderness ratio (λ), calculated based on
in Table 3. The mean value of the PFE /PEC4 ratio was 1.02 with the
EC4 [27] was 0.56. The longitudinal reinforcement bars were four
corresponding coefficient of variation (COV) of 0.013. Second, the
T13 with ultimate strength (fur ) of 460 MPa and having a concrete
fire loads applied in the tests were used in the calculation of the
cover (us ), based on EC4 [20], of 37.5 mm as detailed in [14]. The
load ratios during the fire. The load ratios varied from 0.28 to 0.40.
transverse reinforcement bars were R6 spaced at a spacing (S) of
Finally, the actual fire resistances of the composite columns were
200 mm.
The proposed experimental fire curve is shown in Fig. 3, which predicted, as shown in Table 3.
is lower than the specified fire curve given in EC1 [23]. The furnace The time–axial displacement relationships were also compared
temperature was slightly lower than the proposed curve which experimentally and numerically for column specimen RCC04,
resulted in a modified fire curve that is also shown in Fig. 3. The as shown in Fig. 5. It is shown that there is generally good
use of the lower fire curve has resulted in longer test periods agreement between the experimental and numerical results. The
as reported in [14] and summarized in Table 2. The test periods experimental and numerical results have shown that the column
exceeded 6 h of heating time for fire tests RCC01–RCC04. Only had initial axial shortening (given negative sign in this study) due
the middle part of 3000 mm for the 3300 mm composite column to the application of the static fire load before heating the columns.
length was heated using the modified experimental fire curve The initial axial shortening predicted by the finite element analysis
in the thermal analysis performed in this study. The load ratios, was 1.03 mm compared with 1.07 mm observed from the tests.
defined in this study as the applied load during fire divided by the After 90 min as predicted numerically (135 min as observed
unfactored design strength at ambient temperature determined experimentally), the column started to develop axial expansion
based on EC4 [27], were calculated and presented in Table 3. (given positive sign in this study) due to heating of the column.
The concrete moisture content was 130 kg/m3 . Further details The maximum axial expansion predicted numerically was 5.21 mm
regarding the fire tests are given in [14]. compared with 6.09 mm observed experimentally, as shown in
Fig. 5. Flexural buckling failure (F ) was predicted experimentally
4. Verification of finite element model for axially restrained after 366 min compared with 376 min numerically.
columns
5. Parametric study
The fire tests conducted in [14] were used to verify the
finite element (FE) model investigating the behaviour of axially A total of 40 axially restrained concrete encased steel composite
restrained concrete encased steel composite columns at elevated columns were analyzed at elevated temperatures in the parametric
250 B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254

(a) Reinforcement bar. (b) Structural steel.

(c) Concrete.

Fig. 4. Time–temperature curves obtained experimentally and numerically for restrained concrete encased steel composite column RCC03.

The columns in the parametric study were divided into 8 groups


(G1–G8). Each group had 6 columns having different axial restraint
ratios 0, 0.2, 0.4, 0.6, 0.8 and 1.0. The unrestrained columns (with
zero restraint ratio) were previously analyzed in [18] and were
included in Table 4 for the purpose of comparison. The first four
groups of columns G1–G4 had the same overall cross-sectional
dimensions (B × D) of 250 × 250 mm, while Groups G5–G8 had
B × D of 300 × 300 mm. The structural steel section used in the
composite columns of G1–G8 was the same UC 152 × 152 × 37
section. Groups G1 and G2 had the same coarse aggregate Granite
but with different load ratios 0.3 and 0.5, respectively. Groups G3
and G4 were identical to G1 and G2, except with different coarse
aggregate Marble. Groups G5–G8 were identical to G1–G4, except
for having different overall cross-section dimensions as mentioned
earlier.
The concrete encased steel composite columns investigated
Fig. 5. Time–axial displacement relationships obtained experimentally and
numerically for restrained concrete encased steel composite column RCC04.
in the parametric study had the same overall column length L
as the tests (3300 mm). It should be noted that the effective
buckling length Le is equal to kL for pin-ended column, where
study. The main parameters investigated were different axial k is the effective length factor which is equal to 1.0 for a pin-
restraint ratios, different load ratios during fire, different coarse ended column. The reinforcement bars in the composite columns
aggregates and different slenderness ratios. The dimensions and of G1–G8 had a yield stress (fyr ) and ultimate stress (fur ) of
material properties of the columns are summarized in Table 4. The 376 and 460 MPa, respectively. The concrete strengths of the
axial restraint ratios (β ) varied from 20%–100% of the axial stiffness columns were kept constant, with the average value used in the
of the composite columns at ambient temperature. The load fire tests [14], of 43 MPa compressive cylinder strength (fc ). The
ratios during the fire (defined as the applied load during the fire structural steel material properties were also kept constant of
divided by the unfactored design strength at ambient temperature 320 MPa yield stress (fys ) and 430 MPa ultimate stress (fus ). The
calculated based on EC4 [27]) were 0.3 and 0.5. The coarse longitudinal reinforcement bars were four T13 having a constant
aggregates were siliceous aggregates Granite and Marble, having cover us of 31.5 mm. The transverse reinforcement bars were
different thermal expansion coefficients. The parametric study R6 spaced at 200 mm. The concrete encased steel composite
included different cross-sectional dimensions, as summarized in columns investigated in the parametric study were heated using
Table 4, which resulted in different slenderness ratios λ than the standard fire curve [23] as shown in Fig. 3. The concrete
calculated based on EC4 [27] of 0.56 and 0.69. moisture content was 130 kg/m3 which is identical to that used
B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254 251

Table 4
Axially restrained concrete encased steel composite columns in the parametric study.

Group Specimen Section Steel section λ β Load ratio Coarse agg. FE EC4
B × D (mm) Fire resis. (min) Failure mode Fire resis. (min)

G1 S1 250 × 250 UC 152 × 152 × 37 0.69 0 0.3 Granite 95 F 90


S2 250 × 250 UC 152 × 152 × 37 0.69 0.2 0.3 Granite 102 F 90
S3 250 × 250 UC 152 × 152 × 37 0.69 0.4 0.3 Granite 113 F 90
S4 250 × 250 UC 152 × 152 × 37 0.69 0.6 0.3 Granite 123 F 90
S5 250 × 250 UC 152 × 152 × 37 0.69 0.8 0.3 Granite 132 F 90
S6 250 × 250 UC 152 × 152 × 37 0.69 1.0 0.3 Granite 140 F 90
G2 S7 250 × 250 UC 152 × 152 × 37 0.69 0 0.5 Granite 57 F 90
S8 250 × 250 UC 152 × 152 × 37 0.69 0.2 0.5 Granite 66 F 90
S9 250 × 250 UC 152 × 152 × 37 0.69 0.4 0.5 Granite 75 F 90
S10 250 × 250 UC 152 × 152 × 37 0.69 0.6 0.5 Granite 84 F 90
S11 250 × 250 UC 152 × 152 × 37 0.69 0.8 0.5 Granite 95 F 90
S12 250 × 250 UC 152 × 152 × 37 0.69 1.0 0.5 Granite 107 F 90
G3 S13 250 × 250 UC 152 × 152 × 37 0.69 0 0.3 Marble 91 F 90
S14 250 × 250 UC 152 × 152 × 37 0.69 0.2 0.3 Marble 99 F 90
S15 250 × 250 UC 152 × 152 × 37 0.69 0.4 0.3 Marble 109 F 90
S16 250 × 250 UC 152 × 152 × 37 0.69 0.6 0.3 Marble 119 F 90
S17 250 × 250 UC 152 × 152 × 37 0.69 0.8 0.3 Marble 127 F 90
S18 250 × 250 UC 152 × 152 × 37 0.69 1.0 0.3 Marble 134 F 90
G4 S19 250 × 250 UC 152 × 152 × 37 0.69 0 0.5 Marble 55 F 90
S20 250 × 250 UC 152 × 152 × 37 0.69 0.2 0.5 Marble 64 F 90
S21 250 × 250 UC 152 × 152 × 37 0.69 0.4 0.5 Marble 73 F 90
S22 250 × 250 UC 152 × 152 × 37 0.69 0.6 0.5 Marble 82 F 90
S23 250 × 250 UC 152 × 152 × 37 0.69 0.8 0.5 Marble 90 F 90
S24 250 × 250 UC 152 × 152 × 37 0.69 1.0 0.5 Marble 98 F 90
G5 S25 300 × 300 UC 152 × 152 × 37 0.56 0 0.3 Granite 135 F 90
S26 300 × 300 UC 152 × 152 × 37 0.56 0.2 0.3 Granite 151 F 90
S27 300 × 300 UC 152 × 152 × 37 0.56 0.4 0.3 Granite 160 F 90
S28 300 × 300 UC 152 × 152 × 37 0.56 0.6 0.3 Granite 161 F 90
S29 300 × 300 UC 152 × 152 × 37 0.56 0.8 0.3 Granite 162 F 90
S30 300 × 300 UC 152 × 152 × 37 0.56 1.0 0.3 Granite 162 F 90
G6 S31 300 × 300 UC 152 × 152 × 37 0.56 0 0.5 Granite 84 F 90
S32 300 × 300 UC 152 × 152 × 37 0.56 0.2 0.5 Granite 101 F 90
S33 300 × 300 UC 152 × 152 × 37 0.56 0.4 0.5 Granite 115 F 90
S34 300 × 300 UC 152 × 152 × 37 0.56 0.6 0.5 Granite 128 F 90
S35 300 × 300 UC 152 × 152 × 37 0.56 0.8 0.5 Granite 140 F 90
S36 300 × 300 UC 152 × 152 × 37 0.56 1.0 0.5 Granite 149 F 90
G7 S37 300 × 300 UC 152 × 152 × 37 0.56 0 0.3 Marble 128 F 90
S38 300 × 300 UC 152 × 152 × 37 0.56 0.2 0.3 Marble 143 F 90
S39 300 × 300 UC 152 × 152 × 37 0.56 0.4 0.3 Marble 156 F 90
S40 300 × 300 UC 152 × 152 × 37 0.56 0.6 0.3 Marble 165 F 90
S41 300 × 300 UC 152 × 152 × 37 0.56 0.8 0.3 Marble 162 F 90
S42 300 × 300 UC 152 × 152 × 37 0.56 1.0 0.3 Marble 164 F 90
G8 S43 300 × 300 UC 152 × 152 × 37 0.56 0 0.5 Marble 80 F 90
S44 300 × 300 UC 152 × 152 × 37 0.56 0.2 0.5 Marble 96 F 90
S45 300 × 300 UC 152 × 152 × 37 0.56 0.4 0.5 Marble 109 F 90
S46 300 × 300 UC 152 × 152 × 37 0.56 0.6 0.5 Marble 120 F 90
S47 300 × 300 UC 152 × 152 × 37 0.56 0.8 0.5 Marble 131 F 90
S48 300 × 300 UC 152 × 152 × 37 0.56 1.0 0.5 Marble 141 F 90

in the fire tests [14]. The thermal expansion coefficients for horizontal displacement in the mid-height due to second order
the concrete made with Granite and Marble coarse aggregates effects. However, bear in mind that the restraint to the composite
were taken as 8.6 × 10−6 /°C and 4 × 10−6 /°C, respectively, as column during the expansion stage must also remain during the
recommended by Klieger and Lamond [29]. contraction stage after the column has buckled, as adopted in
[30,31]. Hence, in this study, the time from the start of heating
6. Discussion and comparison with design codes until point C, where the axial displacement went back to initial
state will be defined as the fire resistance of the concrete encased
Fig. 6 shows a comparison between an unrestrained concrete steel composite column. This is because at point C the column will
encased steel composite column specimen S37 and a restrained
behave similar to an unrestrained column and failure takes place.
column specimen S38. It can be seen that the unrestrained
Table 4 summarizes the fire resistance and failure modes of
column experienced the typical ‘‘runaway’’ failure, while the
restrained column behaved differently. The columns started to the axially restrained concrete encased steel composite columns
buckle at points A and B for restrained and unrestrained columns, investigated in the parametric study. In addition, the axial restraint
respectively. For the restrained column, the time from the start ratio–fire resistance relationships of the columns were plotted in
of heating to point A is known as the expanding zone where Fig. 7. It can be seen from Table 4 and Fig. 7 that the fire resistance
the column is experiencing axial expansion. On the other hand, of the concrete encased steel composite columns (defined in this
the time from point A to point C is known as the contracting study as the time from the start of heating to the point where axial
zone where the column is experiencing axial shortening. The displacements went back to original state) is generally increased
axial shortening is the reflection of the initiation of a large with the increase in the axial restraint ratio. This is attributed to
252 B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254

Fig. 6. Comparison of time–axial displacement relationships for unrestrained and


Fig. 8. Time–axial displacement relationships for Groups G1 and G5 having
axially restrained concrete encased steel composite columns.
different slendernesses (λ) of 0.69 and 0.56, respectively, load ratio of 0.3, coarse
aggregate of Granite and different restraint ratios (β ).

Fig. 9. Time–spring force relationships for Group G1.


Fig. 7. Axial restraint ratio–fire resistance relationships for restrained concrete
encased steel composite columns in the parametric study.

the fact that the restraint to the composite column during the
expansion stage must also remain during the contraction stage
after the column has buckled. The main failure modes predicted
from the finite element analysis was flexural buckling (F ) which
could be accompanied by the concrete spalling failure mode
observed experimentally [14]. It should be noted, as stated
in [32], that the definition of failure in fire is a complex issue.
The study [32] defined the fire resistance of the steel columns
investigated as the first point of observed instability. In this case,
the restraint is assumed to be active only in the loading phase of
the column (when the column is expanding) and its compressive
load increasing. When the column is contracting, the restraint is
assumed not to be effective. Hence, the fire resistance of the axially
restrained steel columns investigated decreased with the increase Fig. 10. Time–internal force relationships for Group G1.
in the restraint ratios, which is different to the findings from this
study. This is due to the different criteria defining column failure column and failure occurs. Similar conclusions are obtained by
at elevated temperatures. plotting the time–axial displacement relationships for (G2 and
The time–axial displacement relationships were plotted in Fig. 8 G6), (G3 and G7) and (G4 and G8) having the same load ratio,
for the columns of Groups G1 and G5 having the same load ratio aggregate type but with different slenderness ratios of 0.69 and
of 0.3 and the same coarse aggregate of Granite but with different 0.56, respectively. The time–spring force and time–internal force
slenderness ratios of 0.69 and 0.56, respectively. It can be seen relationships were plotted in Figs. 9 and 10, respectively, as an
that the axial expansion of the columns (given positive sign in example for the axially restrained composite columns of Group G1.
this study) is reduced with the increase in the axial restraint It can be seen that the higher the restraint ratio, the higher the force
ratio. When the column expands, the spring (representing the in the spring and the increase in internal force of the column.
restraining beam in the test) will be subjected to compression Similar to Fig. 8, Fig. 11 plotted the time–axial displacement
(negative value), as shown in Fig. 9, and the internal force in the relationships for the columns of Groups G1 and G2 having the same
column will increase, as shown in Fig. 10 for columns of Group slenderness ratio of 0.69 and the same coarse aggregate of Granite
G1. By continuing heating the column, the column will experience but with different load ratios of 0.3 and 0.5, respectively. It is
axial shortening which reduces the force in the spring until it shown that the axial expansion of the columns is reduced with the
reaches zero. At this stage the column will behave as unrestrained increase in the axial restraint ratio. However, the increase is clear
B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254 253

resistance using the advanced nonlinear finite element analysis


presented in this study would be more accurate.
The fire resistances of the axially restrained concrete encased
steel composite columns investigated in the parametric study
were compared with the predicted values obtained from the
EC4 [20]. The composite columns had a minimum cross-section
dimension of 250 mm, a minimum concrete cover to reinforcement
of 31.5 mm and a minimum concrete cover to structural steel
of 44.1 mm. Hence, the composite columns should have at least
90 min of fire resistance according to EC4 [20], as shown in Table 4.
It is shown that the EC4 is generally conservative for all the axially
restrained concrete encased steel composite columns, except for
some columns with higher load and slenderness ratios. This is
based on the calculation of fire resistance from the start of heating
Fig. 11. Time–axial displacement relationships for Groups G1 and G2 having to the point where axial displacements went back to its original
slenderness ratio (λ) of 0.69, different load ratios of 0.3 and 0.5, respectively, coarse
state.
aggregate of Granite and different restraint ratios (β ).

7. Conclusions

A nonlinear 3-D finite element model for the analysis of


axially restrained concrete encased steel composite columns at
elevated temperatures has been presented in this paper. The
model has considered the nonlinear material properties of steel,
concrete, longitudinal and lateral reinforcement bars as well as
the effect of concrete confinement at elevated temperatures. The
interface between the steel section and concrete, the longitudinal
and lateral reinforcement bars, and the reinforcement bars and
concrete were also considered allowing the bond behaviour to
be modelled and the different components to retain their profile
during the deformation of the column. The initial overall geometric
imperfection was carefully included in the model. The model has
accounted for the different axial restraint ratios to the columns that
Fig. 12. Time–axial displacement relationships for Groups G5 and G7 having
slenderness ratio of 0.56, load ratio of 0.3, different coarse aggregates Granite and varied from 20% to 100% of the axial stiffness of the columns at
Marble, respectively and different restraint conditions (β ). ambient temperature.
The time–temperature relationships, time–axial displacement
in the columns with a small load ratio of 0.3 than that of higher relationships, deformed shapes at failure, failure modes and
load ratio of 0.5. Similar conclusions are obtained by plotting the composite column fire resistances have been predicted using the
time–axial displacement relationships for (G3 and G4), (G5 and G6) finite element model and compared well against the test results.
and (G7 and G8) having the same slenderness ratio, aggregate type In addition, a parametric study was carried out to investigate the
but with different load ratios of 0.3 and 0.5, respectively. effects on the fire resistance and behaviour of the axially restrained
The time–axial displacement relationships were also plotted concrete encased steel composite columns owing to the change
in Fig. 12 for the columns of Groups G5 and G7 having the same in axial restraint ratios, column slenderness ratios, load ratios and
slenderness ratio of 0.56 and the same load ratio of 0.3 but with coarse aggregate type. It is shown that axially restrained concrete
different coarse aggregates of Granite and Marble, respectively. encased steel composite columns behaved differently in a fire
It can also be seen that the axial expansion of the columns is compared to the unrestrained columns. The typical ‘‘runaway’’
reduced with the increase in the axial restraint ratio. However, the failure was not predicted from the finite element analysis for
increase is clearer in the columns with Granite than that of Marble the restrained columns. It is also shown that the effect of the
due to the higher thermal expansion coefficient of Granite. Similar change in axial restraint ratios on the time–axial displacement
conclusions are obtained by plotting the time–axial displacement relationships is clear for the columns with Granite having high
relationships for (G1 and G3), (G2 and G4) and (G6 and G8) having thermal expansion and for the columns with small load ratios.
the same slenderness ratio, load ratio but with different coarse The fire resistances of the columns obtained from the finite ele-
aggregates of Granite and Marble, respectively. ment analyses were compared with the fire resistances calculated
Current design rules specified in EC4 [20] for concrete encased using the Eurocode 4 for composite columns at elevated tempera-
steel composite columns provide tabulated data for the standard tures. It is shown that, the fire resistance of concrete encased steel
fire resistance. According to [20], composite columns made of composite columns predicted using the present advanced non-
totally encased steel sections may be classified as a function of linear finite element analysis would be more accurate than the
the minimum column overall dimensions B or D, the minimum simplified specification predictions. However, it is shown that the
concrete cover c of the steel section and the minimum axis EC4 is generally conservative for all the axially restrained concrete
distance us of the reinforcing bars. However, the fire resistance of encased steel composite columns, except for some columns with
concrete encased steel composite columns depends on many other higher load and slenderness ratios.
parameters as shown in the parametric study. The parameters
include the restraints to thermal expansion, concrete confinement, Acknowledgement
type of coarse aggregate, load ratios during a fire, initial overall
geometric imperfections, transverse reinforcement bars, material The research work described in this paper was supported by a
nonlinearity of steel, concrete, transverse bars as well as the grant from the Research Grants Council of the Hong Kong Special
interface between these components. Hence the predicted fire Administrative Region, China (Project No. HKU719308E).
254 B. Young, E. Ellobody / Engineering Structures 33 (2011) 245–254

References [17] Ellobody E, Young B. Numerical simulation of concrete encased steel


composite columns. J Constr Steel Res 2011;67:211–22.
[1] Mirza SA, Skrabek BW. Statistical analysis of slender composite beam–column [18] Ellobody E, Young B. Investigation of concrete encased steel composite
strength. J Struct Eng, ASCE 1992;118(5):1312–31. columns at elevated temperatures. Thin-Walled Struct 2010;48(8):597–608.
[2] Mirza SA, Hyttinen V, Hyttinen E. Physical tests and analyses of composite [19] ABAQUS standard user’s manual. Vols. 1, 2 and 3. Version 6.8-1. USA: Hibbitt,
steel–concrete beam–columns. J Struct Eng, ASCE 1996;122(11):1317–26. Karlsson and Sorensen, Inc.; 2008.
[3] El-Tawil S, Deierlein GG. Strength and ductility of concrete encased composite [20] EC4. Eurocode 4: design of composite steel and concrete structures—part
columns. J Struct Eng, ASCE 1999;125(9):1009–19. 1–2: general rules—structural fire design. London (UK): British Standards
[4] Furlong RW. Design of steel-encased concrete beam–columns. J Struct Div, Institution. BS EN 1994-1-2; 2005.
ASCE 1968;94(1):267–81. [21] EC3. Eurocode 3: design of steel structures—part 1–1: general rules and rules
[5] Virdi KS, Dowling PJ. The ultimate strength of composite columns in biaxial for buildings. London (UK): British Standards Institution. BS EN 1993-1-1;
bending. Proc Inst Civ Eng, Part 2 1973;55:251–72. 2005.
[6] Roik K, Bergmann R. Design method for composite columns with unsymmet- [22] EC2. Eurocode 2: design of concrete structures—part 1–1: general rules and
rical cross-sections. J Constr Steel Res 1990;33:153–72. rules for buildings. London (UK): British Standards Institution. BS EN 1992-1-
[7] Kato B. Column curves of steel–concrete composite members. J Constr Steel 1; 2004.
Res 1996;39(2):121–35. [23] EC1. Eurocode 1: actions on structures. General actions. Actions on structures
[8] Munoz PR, Hsu CT. Behaviour of biaxially loaded concrete-encased composite exposed to fire. London (UK): British Standards Institution. BS EN 1991-1-2;
columns. J Struct Eng, ASCE 1997;123(9):1163–71. 2002.
[9] Munoz PR, Hsu CT. Biaxially loaded concrete-encased composite columns: [24] Ellobody E, Young B. Design and behaviour of concrete-filled cold-formed
design equation. J Struct Eng, ASCE 1997;123(12):1576–85. stainless steel tube columns. Eng Struct 2006;28(5):716–28.
[10] Chen CC, Lin NJ. Analytical model for predicting axial capacity and behaviour [25] Ellobody E, Young B. Nonlinear analysis of concrete-filled steel SHS and RHS
of concrete encased steel composite stub columns. J Constr Steel Res 2006;62: columns. Thin-Walled Struct 2006;44(8):919–30.
424–33. [26] Bailey CG, Ellobody E. Whole building behaviour of bonded post-tensioned
[11] Malhotra HL, Stevens RF. Fire resistance of encased steel stanchions. Proc Inst concrete floor plates exposed to fire. Eng Struct 2009;31:1800–10.
Civ Eng, ICE 1964;27:77–97. [27] EC4. Eurocode 4: design of composite steel and concrete structures—part
[12] Huang ZF, Tan KH, Toh WS, Phng GH. Fire resistance of composite columns 1–1: general rules and rules for buildings. London (UK): British Standards
with embedded I-section steel-effects of section size and load level. J Constr Institution. BS EN 1994-1-1; 2004.
Steel Res 2008;64:312–25. [28] EC2. Eurocode 2: design of concrete structures—part 1–1: general rules.
[13] Eugene PGH. Fire resistance of steel and composite columns. A thesis Structural fire design. London (UK): British Standards Institution. BS EN 1992-
submitted to the Nanyang Technological University in fulfillment of the 1-2; 2004.
requirement for the degree of master of engineering. 2005. [29] Klieger P, Lamond JF. Significance of tests and properties of concrete and
[14] Huang ZF, Tan KH, Phng GH. Axial restraint effects on the fire resistance concrete making materials. ASTM Publication; 1994.
of composite columns encasing I-section steel. J Constr Steel Res 2007;63: [30] Wang YC. Postbuckling behavior of axially restrained and axially loaded steel
437–47. columns under fire conditions. J Struct Eng, ASCE 2004;130(3):371–80.
[15] Franssen JM, Kodur VKR, Mason J. User’s manual for SAFIR2001—a computer [31] Franssen JM. Failure temperature of a system comprising a restrained column
program for analysis of structures submitted to the fire. Belgium: University submitted to fire. Fire Saf J 2000;34:191–207.
of Liege; 2000. [32] Ali FA, Shepherd P, Randall M, Simms IW, O’Connor DJ, Burgess IW. The effect
[16] Huang ZF, Tan KH. FE simulation of space steel frames in fire with warping of axial restraint on the fire resistance of steel columns. J Constr Steel Res 1998;
effect. Int J Adv Steel Constr 2007;3(3):652–67. 46(1–3): Paper no. 177.

You might also like